基于GPU超级计算系统的天文科学应用  

Applications of the GPU Supercomputing System in Astronomical Scientific Research

作　　者：郑裕民[1] 高娜[1] 刘梁[1] 

机构地区：[1]中国科学院紫金山天文台,南京210008

出　　处：《科研信息化技术与应用》2011年第6期144-152,共9页E-science Technology & Application

摘　　要：天文学主要通过观测来收集天体的各种信息。现代天文学已经成为计算密集型学科,无论是海量观测数据处理,还是理论模型计算和数值模拟,都需要强大的计算能力支撑。GPU高性能、并行的通用计算架构的出现使得改变这种情况成为现实。本文将介绍中国科学院紫金山天文台基于GPU超级计算系统开发的图像处理与分析、数值模拟方法,及其在移动天体目标提取、CCD图像数据处理、暗物质探测以及过渡盘能量分布、行星吸积、物理光学并行软件等方面取得的应用成果。Observation is the most widely used way to collect the information of astronomical objects. Modern astronomy is becoming a computing-intensive science which needs powerful computational capability, no matter for massive data processing, or for theoretical model computation and numerical simulation. The appearance of GPU （Graphic Processing Unit） with its high performance, general parallel computation architecture has made it possible to change this situation. Based on GPU technology, the Purple Mountain Observatory has developed image processing and analysis methods, numerical simulation methods. This paper introduced these methods and their applications in target detection of mobile objects, CCD image data processing, dark matter detection and the energy distribution of the transition plate, planetary absorption product, physical optics parallel software ,etc.

关 键 词：信息化 并行计算 高性能计算 数据处理 数值模拟 

分 类 号：P1[天文地球—天文学]